The present disclosure relates to devices and methods for determining a characteristic of a sample using one or more integrated computational elements, and, more specifically, to devices and methods that computationally combine detector output signals produced from electromagnetic radiation that has optically interacted with one or more integrated computational elements.
Spectroscopic techniques for measuring various characteristics of materials are well known and routinely used under laboratory conditions. In cases where there is not extensive sample matrix interference, spectroscopic techniques can sometimes be carried out without using an involved sample preparation. It is more common, however, to carry out various sample preparation procedures before conducting the analysis. Reasons for conducting sample preparation procedures before carrying out an analysis can include, for example, removing interfering background materials from the sample, converting an analyte of interest into a chemical form that can be better detected by a chosen spectroscopic technique, concentrating a low concentration analyte, and/or adding standards to improve the accuracy of quantitative measurements. Sample preparation steps can result in delays of hours to days when obtaining an analysis. Furthermore, there can be additional delays associated with transporting the sample to a laboratory equipped to carry out the analysis.
Although spectroscopic techniques can, at least in principle, be conducted at a job site in real-time or near real-time, the transitioning of spectroscopic instruments from the laboratory into a field or process environment can be expensive and complex. For example, conditions such as inconsistent temperature, humidity, and vibration can be commonly encountered during field or process use, and they can be difficult to compensate for with conventional spectroscopic instruments. At a minimum, these conditions and others can affect the calibration and durability of many types of spectroscopic instruments. Further, field personnel may not have the training needed to satisfactorily carry out a spectroscopic analysis and take appropriate action in response.
When conducting conventional spectroscopic analyses, low detector signals for a characteristic of interest can sometimes be addressed by determining a derivative of the original spectrum. Obtaining the first derivative (slope) or second derivative (concavity) of the original spectrum may enhance small or subtle spectral features such that they can be more readily discerned. A key disadvantage associated with conventional derivative spectroscopy techniques is that spectral noise is also magnified when a derivative is determined, often making the derivative spectrum exceedingly noisy.
As an alternative to conventional spectroscopic techniques, optical computing devices containing an integrated computational element can be configured to specifically detect a characteristic of interest in a sample. Optical computing devices may utilize electromagnetic radiation to perform calculations, as opposed to the hardwired circuits of conventional electronic processors. Because optical computing devices can be specifically configured to detect a characteristic of interest, there may sometimes be a reduced need to conduct involved sample preparation steps prior to conducting an analysis. Further, optical computing devices are generally operationally simple and rugged. Thus, optical computing devices may be less impacted by the conditions that degrade the performance of conventional spectroscopic instruments, thereby making them well suited for field or process environments.
Unlike conventional spectroscopic analyses, which produce a spectrum that needs to undergo further interpretation to obtain a result, the ultimate output of an optical computing device is a real number that can be correlated with a characteristic of a sample. Correlation of the output of an optical computing device to a sample characteristic may be conducted, for example, by comparing the device's output for a sample against the device's output for one or more standards having a known value of a characteristic of interest. The output simplicity of optical computing devices is one of their more desirable features, which allows them to be deployed with little or no operator training.